Differential Roles of ERK and Akt Pathways in Regulation of EGFR-Mediated Signaling and Motility in Prostate Cancer Cells

Oncogene (2010) 29, 4947–4958 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 www.nature.com/onc ORIGINAL ARTICLE Differential roles of ERK and Akt pathways in regulation of EGFR-mediated signaling and motility in prostate cancer cells

YGan1, C Shi1,3, L Inge2, M Hibner1, J Balducci1 and Y Huang1

1Department of Obstetrics and Gynecology, St Joseph’s Hospital and Medical Center, Phoenix, AZ, USA; 2Heart and Lung Institute, St Joseph’s Hospital and Medical Center, Phoenix, AZ, USA and 3Department of Pathogen Biology, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, PR China

Upregulation of epidermal growth factor receptor (EGFR) Oncogene (2010) 29, 4947–4958; doi:10.1038/onc.2010.240; and subsequent increases in extracellular-regulated kinase published online 21 June 2010 (ERK) and Akt signaling are implicated in prostate cancer progression. Impaired endocytic downregulation of EGFR Keywords: EGFR; ERK; Akt; phosphorylation; signal- also contributes to oncogenic phenotypes such as metastasis. ing; cell migration Thus, understanding the roles of divergent signaling pathways in the regulation of EGFR trafficking and EGFR- driven invasive migration may enable the development of Introduction more effective therapies. In this study, we use the human prostate cancer cell lines, DU145 and PC3, to investigate The molecular mechanisms of prostate cancer are still the effects of both the ERK and Akt pathways on epidermal poorly understood, despite the threat that prostate growth factor (EGF)-mediated EGFR signaling, trafficking cancer poses to the health of men worldwide. As and cell motility. We show that DU145 and PC3 cells prostate tumors are initially dependent on androgens overexpress EGFR and migrate in a ligand (EGF)- for growth and survival, androgen deprivation therapy dependent manner. Next, we show that pharmacological is commonly the first-line treatment for prostate cancer inhibition of ERK (but not Akt) signaling enhances EGF- patients. However, prostate cancer cells can develop a induced EGFR activation, ubiquitination and downregula- hormonal-refractory (androgen independent) state, ren- tion, and may lead to enhanced receptor turnover. These dering principal treatment options palliative because of findings negatively correlate with ERK-mediated threonine the acquisition of invasive and metastatic capacities phosphorylation of EGFR, implicating it as a possible associated with androgen independence. To date, no mechanism. Further, we uncover that EGF promotes effective therapy allows the abrogation of prostate disassembly of cell–cell junctions, downregulation of cancer’s progression to advanced, invasive forms. E-cadherin and upregulation of the transcriptional Recent evidence suggests that acquisition of androgen repressor, Snail, typical characteristics of epithelial–me- independence may be due to upregulation of growth senchymal transition (EMT). These effects are dependent factor/receptor signaling pathways, principally the on activation of Akt, as inhibition of Akt signaling abolishes epidermal growth factor receptor (EGFR) (Kambham- EGF/EGFR-driven cell migration and EMT. Knockdown pati et al., 2005; Traish and Morgentaler, 2009), making of endogenous Snail also prevents EGFR-mediated down- EGFR an attractive target for therapeutic intervention. regulation of E-cadherin, EMT and cell migration. However, the exact contribution that EGFR makes to Surprisingly, inhibition of the ERK pathway augments prostate cancer progression remains unclear. EGFR-dependent motility, occurring concomitantly with EGFR regulates cell growth, differentiation, motility, elevation of EGF-induced Akt activity. Collectively, our adhesion and tumorigenesis through interaction with its results suggest that EGF-triggered ERK activation has cognate ligand, epidermal growth factor (EGF). EGFR profound feedback on EGFR signaling and trafficking by is the prototype of the ErbB family, which also includes EGFR threonine phosphorylation, and Akt has a pivotal ErbB-2, 3 and 4, and is expressed in nearly all epithelial role in EGFR-mediated cell migration by activating EMT. tissues. EGF engagement activates EGFR’s intrinsic More important, our results also suggest that therapeutic kinase and leads to activation of several downstream targeting of ERK signaling may have undesirable outcomes intracellular signaling pathways, including rat sarcoma– (for example, augmenting EGFR-driven motility). MAPK kinase (MEK)–extracellular-related kinase (ERK) and phosphoinositide 3-kinase (PI3K)–Akt pathways, responsible for a variety of mitogenic, metastatic and other tumor-promoting cellular activities Correspondence: Dr Y Huang, Department of Obstetrics and (Wells, 1999; Grant et al., 2002). On EGF binding, Gynecology, St Joseph’s Hospital and Medical Center, 445 North EGFR undergoes a process of internalization, ubiqui- 5th Street, Suite 110, Phoenix, AZ 85004, USA. E-mail: [email protected] tination and destruction (known as EGFR endo- Received 21 January 2010; revised 3 May 2010; accepted 22 May 2010; cytosis and trafficking), resulting in temporary EGFR published online 21 June 2010 downregulation (Wiley, 2003; Citri and Yarden, 2006). ERK and Akt in EGFR signaling and cell motility Y Gan et al 4948 Impaired endocytic downregulation of EGFR is fre- may have undesirable outcomes (for example, enhancing quently associated with cancer, as it can lead to EGFR-driven prostate cancer cell migration). uncontrolled signaling and thus to oncogenic phenotypes (Grandal and Madshus, 2008; Roepstorff et al., 2008). Further data suggest that signaling can also emanate from the EGFR in the process of postendocytic Results trafficking (Wiley, 2003; Sebastian et al., 2006). Indeed, DU145 and PC3 cells highly express EGFR, but not we have previously shown that phosphorylation of ErbB-2, and are responsive to EGF EGFR at threonine-669 by ERK can influence receptor signaling and trafficking (Huang et al., 2003, 2004, 2006; The well-characterized human prostate cancer cell lines, Li et al., 2008). However, the effects that ERK- DU145 and PC3, are both androgen insensitive (van et al. dependent EGFR phosphorylation has on the quanti- Bokhoven , 2003), making them excellent models to tative and qualitative output from EGFR and cancerous study the consequences of EGF-mediated signaling in behaviors, such as invasive migration, remain poorly hormone-refractory prostate cancer. Compared with an understood. Clinically, upregulation of EGFR and/or androgen-responsive human prostate cancer cell line, et al. ErbB-2 signaling is associated with more aggressive LnCap (van Bokhoven , 2003), EGFR proteins behavior in a broad spectrum of human cancers and were overexpressed in DU145 and PC3, whereas ErbB-2 correlates with poor prognosis (Yarden and Sliwkowski, was mainly expressed in LnCap cells (Supplementary 2001; Mendelsohn and Baselga, 2003). Thus, it is Figure 1a), indicating that the two androgen-indepen- of particular importance to understand the roles of dent cell lines (DU145 and PC3) predominantly divergent downstream pathways in the regulation of expressed EGFR, but not ErbB-2. Phosphorylation of 1068Y is known to correlate with EGFR kinase EGFR trafficking and EGFR-mediated cellular pro- residue activation (Rojas et al., 1996). As shown in Supplemen- cesses, as it will enable the development of more effective and selective therapies. tary Figure 1b, EGF induced phosphorylation (activation) of EGFR and activated ERK and Akt in both Acquisition of migratory properties is a prerequisite DU145 and PC3 cells. Notably, Akt was basally for cancer progression and for invasive migration of tumor cells into surrounding tissue. Within carcinoma activated in PC3 (Supplementary Figure 1b, lane 3), consistent with a previous report that this cell line (cancer of epithelial origin) cells, acquisition of inva- harbors PTEN deletion (Davies et al., 2002). Further, siveness requires a dramatic morphological alteration, EGF markedly promoted prostate cancer cell migration termed epithelial–mesenchymal transition (EMT), as measured by wound closure assays (Supplementary wherein carcinoma cells lose their epithelial character- Figure 1c). Collectively, overexpression of EGFR and istics of cell polarity and cell–cell adhesion and switch to robust activation of ERK and Akt in response to EGF a motile mesenchymal phenotype (Thiery, 2002; Thiery and Sleeman, 2006). Disruption of cell–cell adherens in DU145 and PC3 cells make these cells appealing systems for evaluating the roles of ERK and Akt junctions mediated by E-cadherin (one of the epithelial pathways in EGFR-mediated actions. markers) is considered a crucial step in EMT and downregulation of E-cadherin is common in metastatic carcinomas (Cavallaro and Christofori, 2004). Reduced Inhibition of the ERK pathway enhances EGF-induced E-cadherin expression has been found in high-grade EGFR activation and downregulation prostate cancers and is associated with poor prognosis To understand the functional impact of ERK and Akt (Umbas et al., 1992, 1994), reflective of its critical role in inhibition on EGFR activation and trafficking in tumor progression. It is widely believed that down- prostate cancer cells, we used two pharmacological regulation of E-cadherin occurs through transcriptional inhibitors, PD98059 and LY294002, to specifically repression mediated by the protein, Snail (Cano et al., block the MEK–ERK and PI3K–Akt pathway, respec- 2000; Peinado et al., 2007; Moreno-Bueno et al., 2008). tively (Figure 1b). As expected, EGF induced Accumulating evidence indicates that the EGFR family overall tyrosine phosphorylation of EGFR (Figure 1a, and its downstream signaling pathways (for example, upper panel, lanes 2 and 8). Surprisingly, PD98059 PI3K–Akt and rat sarcoma–MEK–ERK) regulate the treatment enhanced EGF-induced EGFR tyrosine expression of Snail (Lee et al., 2008; Qiao et al., 2008; phosphorylation (lane 4 vs 2, lane 10 vs 8), whereas Hipp et al., 2010), suggesting that pharmacological the effect of LY294002 was minimal (lane 6 vs 2, inhibition of these pathways may prevent the loss of lane 12 vs 8). Immunoblotting with a panel of antibodies E-cadherin function and thereby acquisition of invasive that recognize phosphorylated tyrosine residues on motility. EGFR, including 1068Y, 1045Yand845Y (Huang et al., In this study, we explored the effects of Akt and ERK 2003), affirmed that inhibition of the ERK path- pathways on EGFR-mediated signaling and motility in way indeed augmented EGF-induced EGFR tyrosine human prostate cancer cells. Our novel findings suggest phosphorylation (Figure 1c, lane 4 vs 2), Po0.05 that EGF-triggered ERK activation has profound (Figure 1d). feedback on EGFR signaling and trafficking by EGFR To determine the effects of PD98059 and LY294002 threonine phosphorylation, whereas Akt has a pivotal on EGFR trafficking, we compared EGFR protein role in EGFR-mediated migration. We also present levels in DU145 cells pretreated with vehicle (dimethyl evidence that therapeutic targeting of ERK signaling sulfoxide, DMSO), PD98059 or LY294002 for 1 h,

Oncogene ERK and Akt in EGFR signaling and cell motility Y Gan et al 4949 DU145 PC3 DMSOPD LY

PD LY PD LY EGF (min) 0 5 15 305 15 30 5 15 30 - E - E - E - E - E - E 205 IP: EGFR IB: EGFR IB: pTyr 205 IP: EGFR IB: β-actin IB: EGFR 1234567 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10

PD PD 100 DMSO - E - E - E - E 48 LY294002 IB: pERK PD98059 48 IB: ERK 80

LY LY ** - E - E - E - E 60 75 (% of initial)

IB: pAkt Remaining EGFR ** 75 IB:Akt 40 DU145 PC3 0 5 10 15 20 25 30 EGF treatment (min) PD EGF - EEE - E Figure 2 Inhibition of the ERK pathway augments EGF-induced EGF + PD EGFR downregulation. (a) Serum-starved DU145 cells were IB: pY-1068 2 pretreated with vehicle (DMSO), PD98059 (50 mM) or LY294002 * (50 mM) for 1 h and then stimulated with EGF (10 ng/ml) for IB: pY-1045 * 0–30 min. Protein extracts were subjected to immunoblotting with 1.5 anti-EGFR or anti-b-actin (loading control). (b) Statistical analysis

IB: pY-845 DU145 of pooled data from ﬁve independent experiments indicated that IB: β-actin 1 PD98059 signiﬁcantly enhances EGF-induced EGFR downregulation at 15 and 30 min (**Po0.01). The EGFR mass at each time 0.5 point was normalized to the control (lane 1 in (a), set as 100%) and IB: pY-1068 data are mean±s.e.

β PC3 IB: -actin 0

EGF-induced pEGFR level DU145 PC3 1 2 3 4 Figure 1 Inhibition of the ERK pathway enhances EGF-induced time-dependent loss of EGFR (Figure 2a, upper panel, tyrosine phosphorylation (activation) of EGFR. (a) PD98059, but not LY294002, enhances EGF-induced overall tyrosine phospho- lanes 2–4). The kinetics was not altered by LY294002 rylation of EGFR in both DU145 and PC3 cells. Serum-starved (lanes 8–10 vs lanes 2–4). In contrast, PD98059 led to cells were pretreated with vehicle (dimethyl sulfoxide, DMSO), accelerated ligand-induced EGFR downregulation PD98059 (50 mM) or LY294002 (50 mM) for 1 h before stimulation (lanes 5–7 vs lanes 2–4). Statistical analysis is shown in with vehicle (-) or EGF (10 ng/ml) for 15 min. Protein extracts were Figure 2b. Similar results were obtained in PC3 cells immunoprecipitated with anti-EGFR. Eluted proteins were analyzed by immunoblotting with anti-phosphotyrosine (pTyr) or anti- (data not shown). EGFR antibody, as indicated. (b) Pharmacological inhibition of ERK and Akt pathways. Protein extracts as in (a) were resolved on The ERK pathway affects EGFR trafficking by threonine sodium dodecyl sulfate–polyacrylamide gel electrophoresis and phosphorylation of EGFR immunoblotted with anti-pERK, anti-total ERK, anti-pAkt or anti-total Akt. (c) PD98059 enhances EGF-induced tyrosine Consequent to EGF stimulation, activated EGFR is phosphorylation of EGFR at specific tyrosine residues (1068Y, rapidly ubiquitinated by c-Cbl, an SH2 domain-contain- 1045Y and 845Y). Serum-starved cells were pretreated with vehicle ing ubiquitin ligase that itself becomes phosphorylated (DMSO) or PD98059 (50 mM) for 1 h before stimulation with and binds to EGFR, which promotes postinternaliza- vehicle (-) or EGF (10 ng/ml) for 15 min. Protein extracts were tion of EGFR and sorting to endosomes and lysosomes analyzed by immunoblotting with specific antibodies as indicated. (d) Densitometric and statistical analyses of pooled data from three for degradation (Levkowitz et al., 1998; Joazeiro et al., independent experiments indicate that PD98059 pretreatment 1999; Waterman et al., 1999). In DU145 cells, EGF significantly enhances EGF-induced EGFR tyrosine phosphoryla- stimulation led to EGFR ubiquitination (Figure 3a, tion in both DU145 and PC3 cells. Data are plotted as relative upper panel, lane 2), which was enhanced by PD98059 ± phospho-EGFR levels and are mean s.e. *Po0.05. (lane 4 vs 2), but not by LY294002 (lane 6 vs 2). Enhancement of EGFR ubiquitination by PD98059 was statistically significant (Po0.01; Figure 3b), and followed by exposure to EGF for 0–30 min (Figure 2a). correlated with EGFR downregulation (Figure 2). In the absence of EGF, neither PD98059 nor LY294002 We also evaluated the degree of Cbl tyrosine phospho- caused apparent changes in the EGFR mass (data not rylation and the level of Cbl proteins associated shown). As expected, EGF treatment alone resulted in a with EGFR under these conditions (Figures 3c–e) by

Oncogene ERK and Akt in EGFR signaling and cell motility Y Gan et al 4950 NS PD LY 1.5 ** - E - E - E 205 IP: EGFR 1 IB: Ubiquitin EGFR-Ub

205 0.5 IP: EGFR EGFR EGF-induced IB: EGFR EGFR ubiquitination 0 1 2 3 4 5 6 E E+PD E+LY

- E - E 2 ** ** 2 IP: EGFR Cbl IB: Cbl 1.5 1.5 205 1 IP: EGFR pEGFR 1 IB: pTyr pCbl

Total Cbl level 0.5 100 0.5 (EGFR-associated) IP: EGFR (EGFR-associated)

EGFR Phosphorylation of Cbl IB: EGFR 0 0 E E+PD E E+PD 1 2 3 4 Figure 3 Inhibition of the ERK pathway enhances EGF-induced EGFR ubiquitination and association between phosphorylated EGFR and Cbl. (a) Serum-starved DU145 cells were pretreated with vehicle (DMSO), PD98059 (50 mM) or LY294002 (50 mM) for 1 h before stimulation with vehicle (-) or EGF (10 ng/ml) for 15 min. Protein extracts were immunoprecipitated with anti-EGFR, followed by immunoblotting with anti-ubiquitin or anti-EGFR, as indicated. (b) Pooled data from three such experiments, as shown in (a), were subjected to densitometric analysis. Data are plotted as relative levels of EGF-induced EGFR ubiquitination and are mean±s.e. **Po0.01; NS, not statistically signiﬁcant. (c) Serum-starved DU145 cells were pretreated with vehicle (DMSO) or PD98059 (50 mM) for 1 h before stimulation with vehicle (-) or EGF (10 ng/ml) for 15 min. Protein extracts were immunoprecipitated with anti-EGFR, followed by immunoblotting with anti-Cbl, anti-phosphotyrosine (pTyr) or anti-EGFR, as indicated. (d, e) Statistical analyses of pooled data from three independent experiments, as shown in (c), indicate that PD98059 signiﬁcantly increases the EGF-induced levels of both total Cbl and phospho-Cbl associated with EGFR. Data are mean±s.e. **Po0.01.

coimmunoprecipitation experiments. As shown in ERK activity dependent. It is interesting that this Figure 3c, EGF induced physical association of tyrosine threonine phosphorylation negatively correlated with phosphorylated Cbl with activated EGFR, which was the degree of tyrosine phosphorylation, downregulation significantly enhanced by PD98059 (Figure 3c, top and and ubiquitination of EGFR, and the amount of Cbl middle panels, lane 4 vs 2), Po0.01 (Figures 3d and e). proteins associated with EGFR (comparing Figure 4 Similar results were obtained in PC3 cells (Supplemen- with Figures 1–3). Similar results were obtained with tary Figure 2). a separate MEK–ERK pathway inhibitor, UO126 In a reconstitution cell system, we have recently (Figures 4b and c). To confirm that these observations shown that elimination of EGFR phosphorylation were due to inhibition of ERK and not to an off-target at 669T by a point mutation (threonine to alanine) effect of the PD98059 and UO126 compounds, we resulted in accelerated ligand-induced receptor loss introduced a constitutively active MEK1 (c.a. MEK1) (Li et al., 2008). To determine whether the effect of into DU145 cells and assessed its impact on EGFR. PD98059 on EGF-induced activation, ubiquitination As shown in Supplementary Figure 3, forced expression (by Cbl) and downregulation of EGFR was related to of c.a. MEK1 in DU145 cells enhanced EGFR mass and receptor threonine phosphorylation in prostate cancer promoted EGFR threonine phosphorylation, when cells that endogenously express EGFR, we used a compared with the cells expressing lacZ (control). Our monoclonal antibody (PTP101) that specifically recog- previous work using other cell systems (Huang et al., nizes phosphorylated threonine residues residing within 2003, 2006; Li et al., 2008) suggests that EGF-induced the ERK substrate consensus sites (Huang et al., 2003, ERK-mediated threonine phosphorylation of EGFR 2004, 2006; Li et al., 2008). As shown in Figure 4a, EGF could serve as a brake on EGFR downregulation. Taken promoted PTP101-reactive threonine phosphorylation together, the current data indicate that inhibition of of EGFR in DU145 cells (Figure 4a, upper panel, lane ERK signaling eliminates EGF-induced EGFR threo- 2). PD98059, but not LY294002, abolished such nine phosphorylation and leads to the release of the phosphorylation (lane 4 vs 2 and 6), suggesting that ‘brake’, which in turn enhances receptor activation, EGF-induced EGFR threonine phosphorylation was ubiquitination (by Cbl) and downregulation.

Oncogene ERK and Akt in EGFR signaling and cell motility Y Gan et al 4951 PD LY UO126

- E - E - E - E - E UO126 IP: EGFR IP: EGFR - E - E IB: PTP101 IB: PTP101 IB: pY-1068 IP: EGFR IP: EGFR IB: EGFR IB: EGFR IB: EGFR IB: pERK IB: pERK

IB: β-actin IB: ERK IB: ERK 142 3 162 3 4 5 142 3 Figure 4 EGF-induced EGFR threonine phosphorylation is ERK activation dependent. (a) Serum-starved DU145 cells were pretreated with vehicle (DMSO), PD98059 (50 mM) or LY294002 (50 mM) for 1 h before stimulation with vehicle (-) or EGF (10 ng/ml) for 15 min. Protein extracts were either immunoprecipitated with anti-EGFR, followed by immunoblotting with PTP101 (speciﬁcally recognizing phosphorylated threonine residues residing within the ERK substrate consensus sites) or anti-EGFR, or directly immunoblotted with anti-pERK or anti-total ERK, as indicated. (b, c) Serum-starved DU145 cells were pretreated with vehicle (DMSO) or with UO126 (10 mM) for 1 h before stimulation with vehicle (-) or EGF (10 ng/ml) for 15 min. Protein extracts were either immunoprecipitated with anti-EGFR, followed by immunoblotting with PTP101 or anti-EGFR, or directly immunoblotted with anti-pERK, anti-total ERK, anti-phospho-EGFR (pY-1068), anti-EGFR or anti- b-actin, as indicated.

Akt activation is critical for EGFR-mediated prostate Inhibition of the Akt pathway suppresses EGF-mediated cancer cell migration cell migration by blocking Snail upregulation and thereby To understand whether blockade of the ERK or Akt E-cadherin downregulation pathway might have a functional impact on EGFR- Because of the necessary role of Akt in prostate cancer driven migration (Supplementary Figure 1c), we first cell migration, we examined whether EGF-induced performed wound closure assays. As shown in activation of Akt promoted EMT-like morphological Figure 5a, LY294002 markedly inhibited EGF-induced changes in prostate cancer cells. As shown in Figure 6a, cell migration, compared with vehicle (DMSO). Statis- in the absence of EGF, both DU145 and PC3 cells were tical analysis revealed that LY294002 slowed EGF- well organized, tightly packed and formed clustered, mediated cell migration by 43% (Po0.01) (Figure 5b). cobblestone-like structures, typical of epithelial cells and Surprisingly, PD98059 slightly enhanced the ligand- suggestive of strong cell–cell adhesion. EGF treatment driven migration compared with control (Po0.05; resulted in mesenchymal morphological features, speci- Figure 5b). Transwell assays confirmed the augmented fically cell shape elongation and scattering, suggestive of effects of PD98059 and inhibitory effects of LY294002 reduced cell–cell adherens junctions and increased cell on EGF-mediated motility (Figure 5c). Furthermore, we motility. Interestingly, such EGF-induced EMT was found that PD98059 boosted EGF-induced Akt activa- markedly diminished when cells were exposed to tion (Figure 5d, pAkt panel, lane 4 vs 2; lane 10 vs 8) by LY294002, suggesting that activation of Akt is required B23 and 36% in DU145 and PC3 cells, respectively for EGF-driven EMT. (Po0.01; Figure 5e), as did UO126 (Supplementary As downregulation of E-cadherin is a hallmark of Figure 4). As a complementary approach to the EMT, we next assessed the cellular localization and pharmacological inhibition of the P13K–Akt pathway, expression levels of E-cadherin in untreated (control) we expressed constitutively activated (myristoylated) and EGF-treated cells in the presence or absence of Akt (Myr-Akt) in DU145 cells and observed a reverse LY294002. Immunofluorescent staining showed typical effect (Supplementary Figure 5), that is, cells expressing E-cadherin localization at cell–cell junctions in control Myr-Akt migrated much faster than vector control cells cells, but such a staining pattern largely disappeared in (Supplementary Figures 5b and c). In contrast to its EGF-treated cells (Figure 6b), which was consistent with inhibitory effect in control cells, LY294002 had no EGF-induced cell morphological changes (Figure 6a). apparent effect on motility in Myr-Akt-expressing cells Interestingly, LY294002 substantially restored the cell– (Supplementary Figure 5d), because of the chronic cell junction presentation of E-cadherin in the presence activation of Akt (Supplementary Figure 5a, middle of EGF (Figure 6b, EGF þ LY vs EGF). Furthermore, panel, lane 2). Taken together, our data strongly suggest immunoblotting confirmed that EGF caused down- that Akt has a central role in EGF/EGFR-directed regulation of E-cadherin (Figure 7a, upper panel, prostate cancer cell migration, and augmentation of lane 2 vs 1). In contrast, vimentin (a mesenchymal EGF-induced Akt activity by PD98059 (and UO126) marker) was upregulated (Figure 7a, middle panel, lane could contribute to enhanced cell motility. 2 vs 1). LY294002 abolished such differences between

Oncogene ERK and Akt in EGFR signaling and cell motility Y Gan et al 4952 0 h 9 h Wound Closure Assay Control Con EGF

** 60 * DMSO 50 EGF 40 30 20 10 Con % Wound closure 0 PD LY

PD98059 Transwell Assay EGF ** 20 **

15 Con 10

cells per field 5 (EGF-induced) Numbers of migrating

LY294002 0 EGF PD LY

DU145 PC3 EGF PD LY PD LY EGF + PD ** - E - E - E - E - E - E 1.5 ** IB: pERK 1 IB: ERK

0.5 EGF-induced

IB: pAkt Akt activation

IB: Akt 0 1 2 3 4 5 6 7 8 9 10 11 12 DU145 PC3 Figure 5 Effects of ERK and Akt on EGF-induced cell migration. (a) Wound closure assays in DU145 cells. Images were captured at 0 and 9 h after wounding in serum-free media without (control) or with EGF (10 ng/ml) in the presence of vehicle (DMSO), PD98059 (25 mM) or LY294002 (25 mM). Total magniﬁcation: 100. (b) Percentages of wound closure at 9 h under each condition as in (a) are plotted, in which the wound width was normalized to the initial at 0 h (n ¼ 15 per condition). (c) Transwell assays. PC3 cells were seeded in top chambers that were separated by membranes of 8-mm pore size from lower chambers in serum-free media with or without EGF (10 ng/ml) in the presence of vehicle (DMSO), PD98059 (25 mM) or LY294002 (25 mM). The cells were allowed to migrate for 8 h. The number of cells that migrated across the membranes per imaging ﬁeld was counted (n ¼ 12 per condition). (d) Inhibition of the ERK pathway boosts EGF-induced Akt activation. Serum-starved cells were pretreated with vehicle (DMSO), PD98059 (50 mM)or LY294002 (50 mM) for 1 h before stimulation with vehicle (-) or EGF (10 ng/ml) for 15 min. Protein extracts were analyzed by immunoblotting with anti-pERK, anti-total ERK, anti-pAkt and anti-total Akt, respectively. (e) Statistical analysis of pooled data, as shown in (d), indicates that the increases of Akt activity by PD98059 are B 23 and 36% in DU145 (n ¼ 4) and PC-3 (n ¼ 3), respectively. Data are mean±s.e. *Po0.05; **Po0.01.

EGF-treated and untreated cells (Figure 7a, lanes 3 and Snail is one of the several transcriptional factors that 4 vs 1 and 2; Figure 7b). Collectively, these results can suppress E-cadherin gene expression (Batlle et al., suggest that EGF-induced activation of Akt drives EMT 2000; Cano et al., 2000). Stability of the Snail protein is within prostate cancer cells. regulated by GSK3b, a downstream effector of Akt

Oncogene ERK and Akt in EGFR signaling and cell motility Y Gan et al 4953 -LYEGF EGF + LY DU145 PC3

-LYEGF EGF + LY

Immunostaining: E-cadherin / DAPI Figure 6 Effect of inhibition of the Akt pathway on EGF-induced EMT-like morphological changes and loss of E-cadherin expression at cell–cell adhesion. (a) DU145 and PC3 cells were grown for 2 days to allow the formation of clustered epithelial islands, starved, pretreated with vehicle (DMSO) or LY294002 (25 mM) for 1 h and treated with vehicle or EGF (10 ng/ml) for 24 h. Images were captured. Total magnification: 100. (b) LY294002 rescues the EGF-induced loss of E-cadherin at cell–cell adherens junctions. DU145 cells as in (a) grown on coverslips were fixed, subjected to double staining with anti-E-cadherin (green) and DAPI (blue) and examined by confocal microscopy. Total magnification: 630.

(Zhou et al., 2004). In DU145 cells, EGF induced robust Snail knockdown (Supplementary Figure 6). Similar GSK3b phosphorylation (inactivation) and LY294002 results were obtained when Snail siRNA no. 6 was used markedly inhibited this phosphorylation (Figure 7c, (data not shown), ruling out an off-target effect. upper panel), which correlated with Akt activity (middle Inhibition of EGF-induced EMT and cell migration by panel). Consistent with Akt-mediated inactivation of Snail knockdown was consistent with the effects of GSK3b, on EGF stimulation, Snail was upregulated LY294002 (compare Figure 8 with Figures 5–7), (Figure 7d, upper panel, lane 2 vs 1). Interestingly, suggesting a critical role of Snail in these processes. Snail expression levels were reduced to basal in the Taken together, these findings implicate Snail as a presence of LY294002 (Figure 7d, upper panel, lanes 3 central effector of EMT and motility mediated by and 4 vs 1 and 2; Figure 7e), presumably by inactivation EGFR-activated Akt within prostate cancer cells. of Akt and restoration of GSK3b activity (Figure 7c). Similar results were obtained in PC3 cells (data not shown). To confirm the essential role of Snail in EGF-driven Discussion EMT and cell migration, we used small interfering RNAs (siRNAs) to knock down endogenous Snail During tumor progression, cancer cells acquire the expression in DU145 cells. Transfection of Snail siRNA ability to invade surrounding tissue and metastasize to (sequence no. 1 or no. 6) into these cells resulted in distant sites, and these malignant properties are severe reduction in Snail protein levels (Figures 8a and b). reflective of aberrant cell signaling (Martin, 2003). Compared with nonspecific siRNA (control), knock- Overexpression of EGFR and alterations in its down- down of Snail with siRNA no. 1 significantly prevented stream signaling pathways, such as rat sarcoma–MEK– the EGF-induced loss of E-cadherin expression (Figures ERK and PI3K–Akt, are linked to more aggressive 8c and d), and concomitantly suppressed EMT behavior and correlate with poor prognosis of various (Figure 8e), which correlated with a decrease in cell human malignancies, including prostate cancer (Yarden motility (Figure 8f). However, the EGF-induced phos- and Sliwkowski, 2001; Mendelsohn and Baselga, 2003). phorylation of Akt and GSK3b was not attenuated by However, the exact roles of the ERK and Akt pathways

Oncogene ERK and Akt in EGFR signaling and cell motility Y Gan et al 4954 Control LY EGF * NS -E-E 1.5 ** IB: E-cadherin NS 109 1.0

IB: Vimentin 48 0.5

48 Expression level IB: β-actin 0 1 2 3 4 LY LY E-cadherin Vimentin

LY LY 150 * - E - E NS - E - E IB: pGSK3β 100 48 IB: Snail Control 75 32 IB: pAkt 50 48 EGF

IB: ERK Snail level (%) 48 IB: β-actin 0 1 2 3 4 142 3 LY Figure 7 Akt signaling contributes to EGF-driven EMT through the route of Akt-GSK3b-Snail-E-cadherin. (a, b) LY294002 abolishes EGF-induced downregulation of E-cadherin and upregulation of vimentin. DU145 cells were allowed to grow for 2 days, starved, pretreated with vehicle (DMSO) or LY294002 (25 mM) for 1 h and treated with vehicle (-) or EGF (10 ng/ml) for 24 h. Total protein extracts were analyzed by immunoblotting with anti-E-cadherin, anti-vimentin and anti-b-actin, respectively (a). Densitometric analyses were performed using pooled data from four such experiments (b). (c) LY294002 prevents EGF-induced phosphorylation (inactivation) of GSK3b by Akt inhibition. The cells as in (a) were stimulated with vehicle (-) or EGF (10 ng/ml) for 15 min. Protein extracts were analyzed by immunoblotting with anti-pAkt, anti-pGSK3b and anti-b-actin. (d) LY294002 reverses EGF-induced upregulation of Snail. Nuclear proteins were extracted from DU145 cells under the conditions as in (a) and subjected to immunoblotting with anti-Snail antibody. Reprobing the blot with anti-ERK veriﬁed equal protein loadings. (e) Densitometric analysis was performed using pooled data from three such experiments. Data are mean±s.e. *Po0.05; **Po0.01; NS, not statistically signiﬁcant.

in the regulation of EGFR fate (trafficking) and EGFR- downregulation of T669A mutant was much more driven cell motility in prostate cancer cells have not been rapid than that of wild-type EGFR because of loss of systematically studied yet. phosphorylation at 669T (Li et al., 2008). Interestingly, On EGF binding, EGFR is activated through EGF-induced EGFR tyrosine phosphorylation was phosphorylation at several tyrosine residues located in enhanced in prostate cancer cells treated with MEK1 its cytoplasmic tail, after which EGFR undergoes inhibitors (this study), which is similar to that which endocytosis and is targeted for degradation. The we observed previously in the EGFR T669A mutant ligand-induced downregulation of EGFR is a complex, (Li et al., 2008). Thus, this study shows a pivotal tightly regulated process, and impaired endocytic down- role of ERK (in contrast to Akt) in the modulation of regulation is often associated with malignancy (Grandal EGF/EGFR trafficking in a more biologically and and Madshus, 2008; Roepstorff et al., 2008). Thus, the pathologically relevant cell system, in which all EGFR degree and kinetics of EGFR activity and downregula- signaling elements are expressed endogenously. In tion are believed to be critical modulators of EGF addition, we show that forced expression of constitu- actions. We previously uncovered that, in addition to tively active MEK1 (c.a. MEK1), which chronically tyrosine phosphorylation, EGFR also undergoes threo- activates ERK, results in an elevated EGFR expression nine phosphorylation in response to EGF, which is level (Supplementary Figure 3). One explanation for this ERK activation dependent and can slow the pace of interesting phenomenon is that chronic ERK activation ligand-induced receptor downregulation (Huang et al., could modulate EGFR trafficking through enhanced 2003, 2006; Li et al., 2008). In this study, we found that EGFR threonine phosphorylation, leading to increased blockade of the ERK pathway, but not of the Akt receptor levels. However, we cannot rule out other pathway, eliminates EGF-induced EGFR threonine possibilities, such as direct or indirect regulation of phosphorylation, which in turn potentiates receptor EGFR gene expression by ERK, as overexpression tyrosine phosphorylation (activation) and subsequently of c.a. MEK1 in any cell type could lead to fundamental enhances EGFR endocytic downregulation in prostate changes in cell signaling and behaviors, which warrant cancer cells. In particular, acceleration in EGF-induced further investigation. EGFR loss by pharmacological inhibition of the ERK Ligand-induced EGFR endocytic downregulation pathway in prostate cancer cells resembles the effects of contains multiple steps including sorting to clathrin- EGFR T669A mutant (669T was mutated to alanine) in a coated pits, transporting to early endosomes and CHO reconstitution system, in which the EGF-induced destining for lysomal degradation or recycling back to

Oncogene ERK and Akt in EGFR signaling and cell motility Y Gan et al 4955

100

** 50 **

IB: Snail 32 48 IB: β-actin

Snail level (% of Mock) 0 1 2 3 4 Mock NS #1 #6

NS NS Snail 150 ** siRNA siRNA #1

- E - E 100 Con IB: E-cadherin 50 EGF IB: β-actin

E-Cadherin level (%) 0 1 2 3 4 NS Snail siRNA siRNA #1

Con EGF 12 **

6 NS siRNA 4 (EGF-induced) 2 Migrating cells per field 0 Snail siRNA #1

Figure 8 Knockdown of endogenous Snail prevents EGF-induced E-cadherin loss, EMT and cell migration. (a, b) Knockdown of Snail by siRNAs. DU145 cells in six-well plates were transfected with either 50 nM of Snail siRNAs (sequence no. 1 or no. 6) or nonspecific (NS) siRNA for 48 h. Mock: transfection without siRNAs. Total proteins were extracted in the presence of 2% sodium dodecyl sulfate. The efficacy of knockdown was determined by immunoblotting with anti-Snail (a). Statistical analysis of pooled data from three independent experiments indicates that both siRNA no. 1 and no. 6 significantly reduce the endogenous Snail protein level (b). (c, d) Snail knockdown prevents EGF-mediated E-cadherin loss. DU145 cells were transfected with NS siRNA or Snail siRNA no. 1, allowed to grow for 32 h, starved and then treated with vehicle (-) or EGF (10 ng/ml) for 24 h. Expression of E-cadherin was detected by immunoblotting with anti-E-cadherin (c) and statistical analysis of pooled data from four such experiments is shown (d). (e) Snail knockdown blocks EGF-induced EMT process. DU145 siRNA transfectants as in (c) were imaged with total magnification of 100. (f) Snail knockdown reduces EGF-driven cell migration. Serum-starved DU145 siRNA transfectants as in (c) were subjected to transwell assay, in which EGF (10 ng/ml) was used as an attractant in the lower chambers. The cells were allowed to migrate for 8 h. The number of cells that migrated across the membranes per imaging field was counted (n ¼ 17 per condition). Data are mean±s.e. **Po0.01; NS, not statistically significant. the cell surface (Wiley, 2003). The molecular machinery EGFR, which strongly correlates with the augmented controlling these processes remains poorly understood. ubiquitination of EGFR by PD98059. Emerging evi- It is believed that ubiquitination has a key role in dence suggests that Cbl can bind to EGFR directly at ‘tagging’ EGFR for endocytosis. Cbl, a ring-finger phosphorylated tyrosine residue 1045 (Y1045) or domain E3 ubiquitin ligase, is mainly responsible for indirectly by adaptor protein Grb2, which binds to EGFR ubiquitination (Levkowitz et al., 1998). One of phosphorylated EGFR at Y1068 and Y1086 (Levkowitz our novel findings is that inhibition of ERK activity et al., 1999; Waterman et al., 2002). In this study, we further promotes the association between Cbl and show that blockade of the ERK pathway conversely

Oncogene ERK and Akt in EGFR signaling and cell motility Y Gan et al 4956 enhances the tyrosine phosphorylation of EGFR at EMT (Thiery, 2002; Conacci-Sorrell et al., 2003; Lu multiple sites, including Y1045 and Y1068 (Figure 1c), et al., 2003; Cavallaro and Christofori, 2004; Lo et al., presumably by elimination of EGFR threonine phos- 2007), there remains little consensus about which phorylation (Figure 4). This raises several interesting downstream pathways are necessary to mediate questions, such as through which site(s) or tyrosine EGFR-induced EMT. Studies within carcinoma cell residue(s) within the EGFR cytoplasmic tail is the effect lines from disparate epithelial tissues have supplied of the EGFR threonine phosphorylation exerted; evidence for several different signaling pathways whether Cbl is the sole factor in EGFR ubiquitination (b-catenin-TCF/LEF-1, ERK, STAT3), as well as or there are any other contributors, such as adaptor alternate repressors of E-cadherin transcription (Twist, proteins Grb2 and/or Eps15 (Waterman et al., 2002; Slug) involved in activation of EMT (Conacci-Sorrell Grandal and Madshus, 2008; Roepstorff et al., 2008); et al., 2003; Lu et al., 2003; Lo et al., 2007), reflective of and finally whether two completely different types of the fact that cancer cells can use completely distinct EGFR phosphorylation (tyrosine vs threonine phos- pathways under diverse biological contexts. Our work in phorylation) exist and how they are balanced under prostate cancer cells, in addition to the study in cervical physiological and pathological conditions. Deciphering cancer cells (Lee et al., 2008), adds yet another these mechanisms will require more detailed mutagen- mechanistic pathway by which carcinoma cells undergo esis, in combination with biochemical and cell biological EMT, specifically an EGF/EGFR-mediated EMT path- studies in the future. Nevertheless, given the important way through Akt inhibition of GSK3b, followed by implications of endocytic downregulation of EGFR in stabilization of Snail and downregulation of E-cadherin cancer, we are intrigued by the findings herein that the (EGFR-Akt-GSK3b-Snail-E-cadherin-EMT). ‘on and off’ of the ERK pathway markedly influences Interestingly, we observed that inhibition of ERK EGF-induced EGFR trafficking in prostate cancer cells, signaling rather boosts EGF-induced Akt activation, which are reflected in alterations in receptor activation, subsequently enhancing cell motility (Figure 5 and ubiquitination and subsequent downregulation. Supplementary Figure 4). This phenomenon not only Another important and intriguing finding from our supports the notion that Akt is the key node in EGFR- current study is that pharmacological inhibition of the directed migratory pathways but also boils down to the Akt pathway significantly affects EGF-mediated pros- question of how ERK inactivation affects EGF-induced tate cancer cell migration, which involves EGF-driven Akt activity. On the basis of our data, we believe that EMT-like morphological changes. The role of EMT in one mechanism could be through the feedback of ERK cancer progression has long been appreciated (Thiery, on EGFR phosphorylation. One can envision that 2002; Moreno-Bueno et al., 2008). Downregulation of inhibition of ERK activity eliminates EGFR threonine E-cadherin is the key step toward the invasive phase in phosphorylation, resulting in enhanced EGFR tyrosine cancer and E-cadherin gene expression is mainly phosphorylation (activation) and subsequently augmen- regulated by the Slug/Snail family of transcriptional ted activation of the downstream PI3K–Akt pathway. repressors that bind to E-box sequences in the proximal The other possibility is that there exists potential E-cadherin promoter (Hemavathy et al., 2000). Snail is crosstalk between ERK and Akt, which is independent regulated by GSK3b by direct binding and phospho- of EGFR phosphorylation. If so, one would expect a rylation, and inhibition of GSK3b results in upregula- similar increase in basal (nonligand-induced) Akt tion of Snail and downregulation of E-cadherin (Zhou activation and cell migration when the ERK pathway et al., 2004). This implies that Snail and GSK3b together is inhibited. However, our results do not favor the latter function as a molecular switch for many signaling mechanism, as PD98059 (and UO126) only potentiates pathways leading to EMT, and may provide a new ligand-induced Akt activation and cell motility, but not connection of Akt to E-cadherin. Along this line, we the basal, in our experimental settings (Figure 5 and uncover that in prostate cancer cells, EGF induces Akt Supplementary Figure 4). Nevertheless, our findings and activation, which in turn phosphorylates GSK3b, leads several other studies (Schlessinger and Hall, 2004; Zhou to Snail upregulation and subsequently downregulates et al., 2004; Qiao et al., 2008; Wu et al., 2009; Yang and E-cadherin. More important, we show that inactivation Wolf, 2009) highlight that Akt is a critical modulator, of Akt largely eliminates these EGF-mediated signaling when activated by receptor tyrosine kinases (including events, suggesting that Akt is a central player in EGFR- the EGFR family), in cancer invasion and metastasis, directed cell migration. Similar observations have been which frequently involves transcriptional repression of recently reported in cervical cancer cells (Lee et al., E-cadherin by Snail, the key step during EMT. 2008). In addition, we found that knockdown of In summary, in this study, we uncover the distinct endogenous Snail in prostate cancer cells significantly roles of EGF-mediated ERK and Akt pathways in the interrupts EGF-induced E-cadherin loss, EMT and regulation of EGFR signaling, trafficking and cell motility. Taken together, our findings that, in prostate motility in human prostate cancer cells and provide cancer cells, EGF-mediated Akt signaling affects both potential mechanisms that may shed new light on phenotypic and molecular attributes, typical of EMT, molecular targeting therapeutics. It is of particular provide new insights into the molecular mechanisms of importance that inhibition of the ERK pathway can EGFR-driven tumor progression and metastasis. augment EGF-induced Akt activation and cell migra- Although considerable data support a role of receptor tion. The latter finding implies that targeting ERK tyrosine kinases, including EGFR, in the induction of signaling in some cancers (for example, prostate cancer)

Oncogene ERK and Akt in EGFR signaling and cell motility Y Gan et al 4957 may have undesirable outcomes and deserves further Immunofluorescent staining detailed study. The cells were fixed in 4% paraformaldehyde, permeabilized with 0.25% Triton X-100 in PBS and stained with anti-E- cadherin antibody. After rinsing, the primary antibody was detected with fluorescein isothiocyanate-conjugated goat-anti- Materials and methods mouse IgG and nuclei were labeled with DAPI (0.5 mg/ml), as described previously (Tu et al., 2001; Ma et al., 2009). Detailed methods are given in ‘Supplementary Information’ Coverslips were mounted and fluorescent images were accompanying the paper. visualized and captured with a Zeiss Confocal Laser Scanning Microscope LSM 710 (Cal Zeiss USA). Cells, antibodies and other reagents Human prostate cancer cells (DU145, PC3 and LnCap) were Knockdown of Snail maintained in RPMI 1640 medium containing 10% fetal Predesigned siRNAs specifically targeting human Snail were bovine serum (Mediatech, Manassas, VA, USA). Antibodies purchased from Qiagen (Germantown, MD, USA). Sequence are listed in ‘Supplementary Information’. Recombinant no. 1 is 50-GGUGUGACUAACUAUGCAA-30 (sense); se- human EGF was obtained from Invitrogen (Carlsbad, CA, quence no. 6 is 50-GCGAGCUGCAGGACUCUAA-30 USA). PD98059, UO126 and LY294002 were from Cell (sense). Universal negative (nonspecific) control siRNA was Signaling (Beverly, MA, USA). from Sigma-Aldrich (St Louis, MO, USA). DU145 cells cultured in six-well plates were transfected with 50nM of Cell starvation, inhibitor treatment, stimulation, protein siRNA duplexes using HiPerFect Transfection Reagent extraction, immunoprecipitation, immunoblotting and (Qiagen), according to the manufacturer’s instructions. The densitometry transfectants were used for various assays 48–72 h after These procedures were performed as previously described transfection to allow maximum knockdown of Snail as (Huang et al., 2003, 2006; Ma et al., 2009). For experiments in specified in each experiment. which ubiquitination was assayed, N-ethylmaleimide (5 mM) was added to the lysis buffer to prevent post-lysis deubiqui- Plasmids and forced expression of constitutively active MEK1 tination of proteins (Deng et al., 2007). Immunoblotting and Myr-Akt signals were detected with SuperSignal chemiluminescent DU145 cells expressing c.a. MEK1 and Myr-Akt were substrate (Pierce, Rockford, IL, USA), captured using a generated by lentiviral and retroviral infection, respectively. Kodak 4000 MM molecular imager, and quantified using Plasmid constructs, viral production and infection were Kodak Molecular Imaging Software (Kodak, Rochester, NY, described in details in ‘Supplementary Information’. USA). Statistical analysis Cell migration All statistical data were from multiple experiments or Two methods (wound closure and transwell assay) were used measurements and presented as mean±s.e. The significance to measure cell migration as described elsewhere (Zhang et al., of differences was estimated using unpaired t-test and Po0.05 2002; Ma et al., 2009). Please see ‘Supplementary Information’ was considered significant. for details. Conflict of interest Analysis of phenotypic and molecular attributes of EMT Cells were seeded at low density and grown for 2 days to allow The authors declare no conflict of interest. the formation of clustered epithelial islands. Cells were starved for 16 h, pretreated with LY294002 (25 mM) or vehicle for 1 h and then treated with EGF (10 ng/ml) or vehicle for 24 h. Cell Acknowledgements images were taken with a Zeiss Axiovert 200M inverted microscope (Cal Zeiss USA, Thornwood, NY, USA). The cells We thank Prof. Michael Croft for generously providing were lysed directly in radioimmunoprecipitation assay (RIPA) retroviral expression vector for constitutively active (myris- buffer containing 2% sodium dodecyl sulfate, and whole cell toylated) Akt. This work was supported by a St Joseph’s lysates were used for immunoblotting to detect the expression Foundation Startup Fund (to YH). Part of this work was of E-cadherin, vimentin and Snail, as specified in each presented at the 91st Endocrine Society Annual Meeting in experiment. Washington DC, USA, 2009.

References

Batlle E, Sancho E, Francı´C, Domı´nguez D, Monfar M, Baulida J Citri A, Yarden Y. (2006). EGF-ERBB signalling: towards the systems et al. (2000). The transcription factor snail is a repressor of level. Nat Rev Mol Cell Biol 7: 505–516. E-cadherin gene expression in epithelial tumour cells. Nat Cell Biol Conacci-Sorrell M, Simcha I, Ben-Yedidia T, Blechman J, Savagner P, 2: 84–89. Ben-Ze’ev A. (2003). Autoregulation of E-cadherin expression by Cano A, Pe´rez-Moreno MA, Rodrigo I, Locascio A, Blanco MJ, del cadherin-cadherin interactions: the roles of beta-catenin signaling, Barrio MG et al. (2000). The transcription factor snail controls Slug, and MAPK. J Cell Biol 163: 847–857. epithelial-mesenchymal transitions by repressing E-cadherin expres- Davies MA, Kim SJ, Parikh NU, Dong Z, Bucana CD, Gallick GE. sion. Nat Cell Biol 2: 76–83. (2002). Adenoviral-mediated expression of MMAC/PTEN inhibits Cavallaro U, Christofori G. (2004). Cell adhesion and signalling by proliferation and metastasis of human prostate cancer cells. Clin cadherins and Ig-CAMs in cancer. Nat Rev Cancer 4: 118–132. Cancer Res 8: 1904–1914.

Oncogene ERK and Akt in EGFR signaling and cell motility Y Gan et al 4958 Deng L, He K, Wang X, Yang N, Thangavel C, Jiang J et al. (2007). Moreno-Bueno G, Portillo F, Cano A. (2008). Transcriptional regula- Determinants of growth hormone receptor down-regulation. Mol tion of cell polarity in EMT and cancer. Oncogene 27: 6958–6969. Endocrinol 21: 1537–1551. Peinado H, Olmeda D, Cano A. (2007). Snail, Zeb and bHLH factors Grandal MV, Madshus IH. (2008). Epidermal growth factor receptor in tumour progression: an alliance against the epithelial phenotype? and cancer: control of oncogenic signalling by endocytosis. J Cell Nat Rev Cancer 7: 415–428. Mol Med 12: 1527–1534. Qiao M, Sheng S, Pardee AB. (2008). Metastasis and AKT activation. Grant S, Qiao L, Dent P. (2002). Roles of ErbB family receptor Cell Cycle 7: 2991–2996. tyrosine kinases, and downstream signaling pathways, in the control Roepstorff K, Grvdal L, Grandal M, Lerdrup M, van Deurs B. of cell growth and survival. Front Biosci 7: d376–d389. (2008). Endocytic downregulation of ErbB receptors: mechanisms Hemavathy K, Ashraf SI, Ip YT. (2000). Snail/slug family of and relevance in cancer. Histochem Cell Biol 129: 563–578. repressors: slowly going into the fast lane of development and Rojas M, Yao S, Lin YZ. (1996). Controlling epidermal growth factor cancer. Gene 257: 1–2. (EGF)-stimulated Ras activation in intact cells by a cell-permeable Hipp S, Walch A, Schuster T, Losko S, Laux H, Bolton T et al. (2010). peptide mimicking phosphorylated EGF receptor. J Biol Chem 271: Activation of epidermal growth factor receptor results in Snail 27456–27461. protein but not mRNA over-expression in endometrial cancer. Schlessinger K, Hall A. (2004). GSK-3beta sets Snail’s pace. Nat Cell J Cell Mol Med 13(9B): 3858–3867. Biol 6: 913–915. Huang Y, Chang Y, Wang X, Jiang J, Frank SJ. (2004). Growth Sebastian S, Settleman J, Reshkin S, Azzariti A, Bellizzi A, Paradiso A. hormone alters epidermal growth factor receptor binding affinity (2006). The complexity of targeting EGFR signalling in cancer: from via activation of ERKs in 3T3-F442A cells. Endocrinology 145: expression to turnover. Biochim Biophys Acta 1766: 120–139. 3297–3306. Thiery JP. (2002). Epithelial-mesenchymal transitions in tumour Huang Y, Kim SO, Jiang J, Frank SJ. (2003). Growth hormone- progression. Nat Rev Cancer 2: 442–454. induced phosphorylation of epidermal growth factor (EGF) Thiery JP, Sleeman JP. (2006). Complex networks orchestrate epithelial- receptor in 3T3-F442A cells. Modulation of EGF-induced traffick- mesenchymal transitions. Nat Rev Mol Cell Biol 7: 131–142. ing and signaling. J Biol Chem 278: 18902–18913. Traish AM, Morgentaler A. (2009). Epidermal growth factor receptor Huang Y, Li X, Jiang J, Frank SJ. (2006). Prolactin modulates expression escapes androgen regulation in prostate cancer: phosphorylation, signaling and trafficking of epidermal growth a potential molecular switch for tumour growth. Br J Cancer 101: factor receptor in human T47D breast cancer cells. Oncogene 25: 1949–1956. 7565–7576. Tu Y, Huang Y, Zhang Y, Hua Y, Wu C. (2001). A new focal adhesion Joazeiro CA, Wing SS, Huang H, Leverson JD, Hunter T, Liu YC. protein that interacts with integrin-linked kinase and regulates cell (1999). The tyrosine kinase negative regulator c-Cbl as a RING- adhesion and spreading. J Cell Biol 153: 585–598. type, E2-dependent ubiquitin-protein ligase. Science 286: 309–312. Umbas R, Isaacs WB, Bringuier PP, Schaafsma HE, Karthaus HF, Kambhampati S, Ray G, Sengupta K, Reddy VP, Banerjee SK, Van Oosterhof GO et al. (1994). Decreased E-cadherin expression is Veldhuizen PJ. (2005). Growth factors involved in prostate associated with poor prognosis in patients with prostate cancer. carcinogenesis. Front Biosci 10: 1355–1367. Cancer Res 54: 3929–3933. Lee MY, Chou CY, Tang MJ, Shen MR. (2008). Epithelial- Umbas R, Schalken JA, Aalders TW, Carter BS, Karthaus HF, mesenchymal transition in cervical cancer: correlation with tumor Schaafsma HE et al. (1992). Expression of the cellular adhesion progression, epidermal growth factor receptor overexpression, and molecule E-cadherin is reduced or absent in high-grade prostate snail up-regulation. Clin Cancer Res 14: 4743–4750. cancer. Cancer Res 52: 5104–5109. Levkowitz G, Waterman H, Ettenberg SA, Katz M, Tsygankov AY, van Bokhoven A, Varella-Garcia M, Korch C, Johannes WU, Smith Alroy I et al. (1999). Ubiquitin ligase activity and tyrosine EE, Miller HL et al. (2003). Molecular characterization of human phosphorylation underlie suppression of growth factor signaling prostate carcinoma cell lines. Prostate 57: 205–225. by c-Cbl/Sli-1. Mol Cell 4: 1029–1040. Waterman H, Katz M, Rubin C, Shtiegman K, Lavi S, Elson A et al. Levkowitz G, Waterman H, Zamir E, Kam Z, Oved S, Langdon WY (2002). A mutant EGF-receptor defective in ubiquitylation and et al. (1998). c-Cbl/Sli-1 regulates endocytic sorting and ubiquitina- endocytosis unveils a role for Grb2 in negative signaling. EMBO J tion of the epidermal growth factor receptor. Genes Dev 12: 21: 303–313. 3663–3674. Waterman H, Levkowitz G, Alroy I, Yarden Y. (1999). The RING Li X, Huang Y, Jiang J, Frank SJ. (2008). ERK-dependent threonine finger of c-Cbl mediates desensitization of the epidermal growth phosphorylation of EGF receptor modulates receptor downregula- factor receptor. J Biol Chem 274: 22151–22154. tion and signaling. Cell Signal 20: 2145–2155. Wells A. (1999). EGF receptor. Intl J Biochem Cell Biol 31: 637–643. Lo HW, Hsu SC, Xia W, Cao X, Shih JY, Wei Y et al. (2007). Wiley HS. (2003). Trafficking of the ErbB receptors and its influence Epidermal growth factor receptor cooperates with signal transducer on signaling. Exp Cell Res 284: 78–88. and activator of transcription 3 to induce epithelial-mesenchymal Wu Y, Deng J, Rychahou PG, Qiu S, Evers BM, Zhou BP. (2009). transition in cancer cells via up-regulation of TWIST gene Stabilization of snail by NF-kappaB is required for inflammation- expression. Cancer Res 67: 9066–9076. induced cell migration and invasion. Cancer Cell 15: 416–428. Lu Z, Ghosh S, Wang Z, Hunter T. (2003). Downregulation of Yang CC, Wolf DA. (2009). Inflamed snail speeds metastasis. Cancer caveolin-1 function by EGF leads to the loss of E-cadherin, Cell 15: 355–357. increased transcriptional activity of beta-catenin, and enhanced Yarden Y, Sliwkowski MX. (2001). Untangling the ErbB signalling tumor cell invasion. Cancer Cell 4: 499–515. network. Nat Rev Mol Cell Biol 2: 127–137. Ma F, Zhang D, Yang H, Sun H, Wu W, Gan Y et al. (2009). Zhang Y, Chen K, Guo L, Wu C. (2002). Characterization of Endothelial cell-specific molecule 2 (ECSM2) modulates actin PINCH-2, a new focal adhesion protein that regulates the remodeling and epidermal growth factor receptor signaling. Genes PINCH-1-ILK interaction, cell spreading, and migration. J Biol Cells 14: 281–293. Chem 277: 38328–38338. Martin GS. (2003). Cell signaling and cancer. Cancer Cell 4: 167–174. Zhou BP, Deng J, Xia W, Xu J, Li YM, Gunduz M et al. (2004). Mendelsohn J, Baselga J. (2003). Status of epidermal growth factor Dual regulation of Snail by GSK-3beta-mediated phosphorylation receptor antagonists in the biology and treatment of cancer. J Clin in control of epithelial-mesenchymal transition. Nat Cell Biol 6: Oncol 21: 2787–2799. 931–940.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene